扩展型关联规则和原关联规则及其若干性质

已有的关联规则模型所反映的基本上是频繁事件中所隐藏的一种积极或肯定关系，而没有反应其隐含的否定关系，在实际应用中，这种否定关系与肯定关系一样也是很重要的，在此论述的扩展型关联规则模型就能反映上述两种关系，据此可以得到更多的规则知识，此外，由于Aproiri关联规则生成方法产生的关联规则具有较大的冗余性，论述的原关联规则可以消除关联规则的这种冗余特性，挖掘原关联规则既能大大减少关联规则的数,目,又能节省规则生成时间，把扩展型关联规则和原关联规则相结合，可使得对扩展关联规则的挖掘更加有效，实验结果也支持了上述结论。

THE EXTENDED ASSOCIATION RULES AND ATOM ASSOCIATION RULES

The association rules model presented by Rakesh Agrawal et al reveals the positive relationships behind the frequent items. However it doesn't give the negative relationships behind the frequent items. However, these negative relationships are as important as those positive relationships in practical applications. The model of extended association rules proposed in this paper can reveal both positive and negative relationships behind frequent items so as to mine more rules. Moreover, the association rules produced by Apriori approach include many redundant rules. Hence, the atom association rules are proposed to remove their redundancy, which can reduce the size of association rules and the time to set up rules. Several conclusions about the support of the extended association rules and the atom association rules are also proved, and their functions and meanings are explained. The experimental results show that those results are applicable, and the efficiency to mine extended association rules has been improved by combining extended and atom association rules. In particular, for mining the atom association rules, the number of association rules can be greatly reduced and the running time is saved.